Hacker NewsHN submitter: pretext
Original published:

Combatting Adversarial Distillation

Combatting Adversarial Distillation

OpenAI employs a layered approach to combat adversarial distillation, combining model training, safeguards, detection, and enforcement to protect against capability extraction.

Key points

  • •OpenAI uses a layered approach to address adversarial distillation.
  • •The strategy includes model training, safeguards, detection, and enforcement.
  • •These measures aim to protect against the extraction of AI capabilities.
#Security & Misuse#Major Players#OpenAI
See more